Increasing scale poses additional problems for systems that have even some, if not all of the characteristics of ultra-large-scale (ULS) systems. We adopt a multi-disciplinary approach to the development, evaluation, and evolution of ULS systems. In particular, we focus on the interaction between the social and computational aspects of ULS systems and investigate how those interactions affect achievement of mission goals. Newly created dynamic mechanisms allow us to broaden our application of the theory of computational mechanism design by accounting for human incentives and uncertainty as they evolve.

We continue to enhance our laboratory proof of concept to demonstrate this. Our proof of concept involves allocating a set of unmanned aerial vehicles (UAVs) to a collection of units that require them to carry out their respective missions. Developing the proof of concept requires finding computationally efficient ways of solving the Markov decision processes that arise as part of the mechanism, using and developing resource-allocation techniques for adaptive networks, and using swarm theory to manage the set of UAVs.

ULS systems continuously evolve, not according to the dictates of a central authority, but rather in response to the aggregate influences of many forces of change. We create and analyze new architecture representations that model the dynamics of the social and computational dimensions and also develop multi-agent simulations of ULS system evolution to gain insight into the principles for how such systems evolve.

The SEI is in a unique position to do this research, having had years of experience in developing and using architecture evaluation methods.

Specific tasks under investigation this year are

• ULS systems perspective proof-of-concept—use the allocation of UAVs to a changing set of missions to demonstrate how the social and computational aspects of a ULS system interact
• ULS architecture behavior—create new architecture models for analyzing the behavior and evolution of ULS systems